Fairer Voting Math: Randomly Choosing From a Shortlist Works Surprisingly Well

When groups pick winners—whether in elections, hiring committees, or community votes—there is a hidden question: how good is the final choice at representing everyone? Researchers measure this as "distortion": how far the winning option is from what would be ideal for the whole group. Lower distortion means fairer results. For ordinary voting, the best deterministic method has a distortion of 3. But if you allow randomness, you can get down to 2.5.

This new paper shows you do not need full randomness to get that advantage. A simple rule—first narrow the options to a shortlist, then pick one name out of a hat—can come very close to the golden 2.5. Specifically, if you use a shortlist of at least 164 candidates, you can beat the deterministic limit of 3. And with 802 candidates on the list, you approach the best possible random outcome. That is a big deal because this kind of "bounded randomness" is easier to explain, audit, and trust.

The math behind it is a clever "compression" trick: if you already have a good random plan, you can shrink it to a shortlist without losing much fairness. The authors even give a formula showing exactly how much fairness you lose as the list gets shorter. The loss shrinks quickly as the list grows, meaning even a modest shortlist gives most of the benefit.

So what does this mean for real life? It suggests that adding a little lottery to decision-making—like picking a jury, allocating funding, or selecting finalists—can make outcomes fairer than pure voting, while staying simple enough for people to understand. The main catch: the math assumes voters' preferences can be measured like distances, which works well for issues like location, price, or policy proximity, but less neatly for purely emotional choices.